Evaluating Impacts of Climate Change on Traditional Mexican Maize

• Climate change is expected heavily impact natural and human systems worldwide (Walther, G.-R. et al., 2002; IPCC 2007, 2013; Kang, Y, S. Khan 2009; Hoegh-Guldberg, O. 2010)

• The IPCC reports numerous negative impacts of climate change on domesticated crops regardless of the emission scenario implemented(Brown and Funk 2008; Ureta et al. 2012; IPCC 2013)

• Ensuring global food security in an increasingly warming world with exponentially increasing world population requires multidisciplinary research (Brown and Funk 2008; Ureta et al. 2012; IPCC 2013, 2015)

Five percent reduction in crop season, sensitivity to change, capacity to cope. Source: CGIAR.

• Maize (Zea mays subsp. mays L.) supplies a staple food crop for more than 200 million people worldwide (Nuss and Tanumihardjo 2010; Ureta et al. 2013)

• Climate change is expected to alter maize suitability, especially between the tropics (Ramirez-Cabral et al., 2017)

• Mexico, maize's C.O.D., harbors traditional maize landraces that account for 60% of maize genetic diversity globally (Ureta et al. 2013)

• Preservation of agrobiodiversity at centers of domestication critical for global food security (Thrupp 2000; Esquinas-Alcazar 2005; Ureta et al. 2013).

Source: The International Maize and Wheat Improvement Center (CIMMYT)

Source: https://lacienciaysusdemonios.com

• > 9,000 years from Mexican annual teosinte (Z. mays ssp. Parviglumus and ssp. mexicana ) via artificial selection by indigenous cultures (Matsuoka et al. 2002; Kato et al. 2009)
• Maize landraces have been grown (Ruiz Corral et al. 2008):
  • from sea-level to 2,900 (m asl)
  • avg. growing temp 12.0 °C - 29.1 °C
  • avg. seasonal precip. 400 mm to 3555 mm
• Regional clusters of phenotypically similar landraces exist (Cutler 1942; Anderson 1947; E.J. Wellhausen et al., 1952; Dyer & Lopez-Feldman 2013)

• As a domesticated crop, maize is intimately contingent upon societal influences:
  • Seed management practices (Anderson 1947; Dyer and Lopez-Feldman 2013)
  • Indigenous religious practices (R. Ortega-Paczka., 2003)
  • Ethnolinguistic diversity (Perales, Benz, and Brush 2005; Brush and Perales 2007; Rivero-Romero et al. 2016)
  • Indigenous agricultural knowledge (Garcia-Martinez et al. 2016; Rivero-Romero et al. 2016)
  • Agro-technological practices (Garcia-Martinez et al. 2016; Rivero-Romero et al. 2016; Toledo and Barrera-Bassols 2017)
  • variations across ethnic groups (Perales, Benz, and Brush 2005; Brush and Perales 2007)

• La Comision Nacional para el Conocimiento y Uso de la Biodiversidad (CONABIO)
• 2000 indigenous population % by municipio (n= 2446 + 16)
• 68 indigenous macro-languages in Mexico (Instituto Nacional de Lenguas Indigenas (INALI))
  • In reality, many more distinct languages
  • 10-14% identify as indigenous, but only 6% speak an indigenous language
  • Loss of potentially irrecoverable deep climatic knowledge

Source: (Edmondson, 1971)

• Joseph Grinnell (Grinnell, 1917) - niches corresponded to species' environments
• Charles Elton (Elton, 1927) - associated species niches through biotic relationships
• George Evelyn Hutchinson - species role in environment (Hutchinson, 1957; Colwell and Rangel 2009).
  • The Hutchinsonian Niche - an “n-dimensional hypervolume” in ecological space in which a species can persist (Colwell and Rangel 2009; Holt 2009)
  • Fundmental Niche - A species' potential geographic distribution in true 4D space, constrained by its n-dimensional bioclimatic envelope
  • Realized Niche - fundamental niche constrained by 'bionomic' interactions; the niche that exists in reality and that does not overlap with other species niches (Griesemer 1992,Pulliam 2000)

Source: (Guisan et al., 2017) Credit: (Guisan et al., 2017,A. Townsend Peterson et al. 2011; Soberon & Peterson, 2005)

• G : Studied geographic area
  • A : Suitable abiotic environment (Grinellean niche)
  • B : Suitable biotic environment (Eltonian niche)
  • C : Colonizable range

1. Realize Niche, suitable to all three (Hutchinson's Niche)
2. Sutiable abiotic with unsuitable biotic conditions (e.g.: high competition)
3. Colonization outside of suitable environment (sink populations)
4. Sink in unsuitable biotic and abiotic environments (historical effects)

• Uses species presence and/or absence observation data and ecological data at those observation sites to estimate complex relationships of species niche (Elith et al. 2011).
• Project hyperdimensional ecological space onto 2- or 3D geographic space
• Numerous algorithms developed for and adapted to ENM
• Ensemble Ecological Niche Modelling weights individual models based on performance(Filho et. al, 2010, Araujo & New 2007)
• Reduce uncertainty across model type, model repetition, Pseudo-absence selection, etc.
• WA and Mean consensus methods provided significantly more robust predictions than single-models and the other consensus methods (Marimon et al., 2009)

• 22931 available through La Comision Nacional para el Conocimiento y Uso de la Biodiversidad (CONABIO)
• Cleaned records of erroneous records and those flagged 'Inconsistent'
• 18060 remaining records
• 64 unique maize races

• WorldClim

  • 30 arc-second resolution
  • ANUSPLIN up to 60,000 weather stations (with >10 years data)
  • largest available archive of downscaled climate data for ecology (Guisan et al., 2017)
  • >2,700 citations for ecological studies including ENMs and SDMs (Booth et al., 2014)

• Current Data from WorldClim 2.0 (Fick & Hijmans, 2017)

  • 1970-2000
  • Maximum and Minimum Monthly Temperatures (°C * 10)
  • Monthly Precipitation (mm)
  • also includes solar radiation, windspeed, water vapor pressure

• Future Data from Worldclim 1.4 (Hijmans et al., 2005)

  • 2041-2060, 2061-2080
  • dynamically-downscaled from 19 GCM used in CMIP5 (Coupled Model Intercomparison Project 5; Meehl et al., 2009) and the Fifth Assessment Report (IPCC, 2013 )
  • RCP 8.5

• Derived from first ecological niche modelling alogorithm 'BIOCLIM' (Nix, 1986)
• More ecologically importance predictor variables
• Can reduce multicollineary, VIF; expands options
• Calculated from monthly Tmax, Tmin, Tmean, and Prec
• 19 Bioclimatic proxy variables from monthly means using 'dismo' (Hijmans et al., 2017)
• Potentially 16 'complementary' variables from 'envirem' package in R (Title & Bemmels, 2017)
  • Must use (WorldClim 2.0) current terrestial solar radiation for both current and future climatologies
  • Must average Tmin and Tmax to get Tmean (.99 R²⁾ (Title & Bemmels, 2017)

1970-2000 Bioclimatic Variables

• BIO1 = Annual Mean Temperature
• BIO2 = Mean Diurnal Range (Mean of monthly (max temp - min temp))
• BIO3 = Isothermality (BIO2/BIO7) (* 100)
• BIO4 = Temperature Seasonality (standard deviation *100)
• BIO5 = Max Temperature of Warmest Month
• BIO6 = Min Temperature of Coldest Month
• BIO7 = Temperature Annual Range (BIO5-BIO6)
• BIO8 = Mean Temperature of Wettest Quarter
• BIO9 = Mean Temperature of Driest Quarter
• BIO10 = Mean Temperature of Warmest Quarter

• BIO11 = Mean Temperature of Coldest Quarter
• BIO12 = Annual Precipitation
• BIO13 = Precipitation of Wettest Month
• BIO14 = Precipitation of Driest Month
• BIO15 = Precipitation Seasonality (Coefficient of Variation)
• BIO16 = Precipitation of Wettest Quarter
• BIO17 = Precipitation of Driest Quarter
• BIO18 = Precipitation of Warmest Quarter
• BIO19 = Precipitation of Coldest Quarter

• Annual PET
• Thornthwaite Aridity Index
• Climatic Moisture Index
• Continentality
• EmbergerQ
• Growing Deg Days 0
• Growing Deg Days 5
• Growing Deg Days 10
• Max Temp Coldest
• Min Temp Warmest

• Month Count By Temp 10
• PET Coldest Quarter
• PET Driest Quarter
• PET seasonality
• PET Warmest Quarter
• PET Wettest Quarter
• Thermicity Index

• Terrain Indices from Elevation from WorldClim 1.4 via SRTM (Shuttle Radar Topographic Mission) with 'raster' package (Hijmans, 2017) (Wilson et al., 2007)
  • Aspect
  • Slope
  • Roughness (difference between the maximum and the minimum value of a cell and its 8 surrounding cells)

• Land Cover Classification
• Soil Quality 1-7
• 19 'dismo' bioclimatic variables
• 16 supplemental 'envirem' bioclimatic variables

Method

• 'usdm' R package for vif objects (Naimi et al., 2014)
• preferred over correlation thresholds due to 'hidden' correlation structures (Guisan et al., 2017)
• correlation structures may change over time
• select variables of ecological importance
• 'vifstep' with VIF threshold of 10 to get remaining variables (Guisan et al., 2017)

# using all data, no pre-selection
presvifstep@results

               Variables      VIF
1      BIO8MeanTWettestQ 3.169749
2  BIO13PrecWettestMonth 7.203552
3      BIO15PrecSeas.COV 5.426713
4      BIO18PrecWarmestQ 5.112587
5      BIO19PrecColdestQ 6.479971
6      EVMminTempWarmest 3.258270
7     EVMmonthCountByT10 2.206976
8         EVMPETColdestQ 5.539127
9          EVMPETDriestQ 2.507005
10        EVMPETWarmestQ 4.670706
11               IrrCult 1.169090
12          Rain.fedCult 1.599255
13        Grass.Woodland 2.314950
14                Barren 1.730909
15                 Urban 1.075483
16                 Water 1.107406
17                aspect 1.023113
18                 slope 1.446115